Chris Holmes wrote:
On 3/29/06, Andrea Aime <[EMAIL PROTECTED]> wrote:
Chris Holmes ha scritto:
This is a great project, definitely what is needed to get the
'geospatial web' to take off. Some of our GeoServer users have had
success just sticking GeoServer's WMS behind squid and making standard
requests with ka-map.
Standard requests with ka-map? I thought ka-map was designed specifically
for MapServer. Did it changed lately, or someone made the necessary
modifications
to it? If so, are they available on the net?
A ready to use ka-map front-end from Geoserver would be really great :-)
Saul Farber did the work, and I think he said that the changes to ka-map
were pretty trivial to use WMS instead of the pre-generated tiles. The
nice thing he did, and the necessary step to get it working right, was
to stick squid in front of GeoServer so that it caches the images generated.
I think there may have been a small bug fix to allow GeoServer to set
the headers. We've been neglecting our JIRA to roll patches in, and
Saul has quite a few good ones sitting there. I'm sorry Saul, we'll try
to get to it soon.
His original email is below, perhaps if he has more info we can convince
him to put it up on the wiki. Saul, you could put it up somewhere in:
http://docs.codehaus.org/display/GEOSDOC/User+Experiences
Chris
Saul Farber wrote:
> Hello all,
>
> I've finally gotten our geoserver-based ka-map demo onto the internet.
>
> It can be seen at http://maps.massgis.state.ma.us/~sfarber/ka-map/
>
> Here's a brief rundown.
>
>
> 1) a web-server (maps.massgis.state.ma.us) hosts the ka-map javascript
> and html files. There's an "init.php" file, but it's not actually doing
> anything php-related...just a text file with the right name so that
> ka-map picks it up.
>
> 2) a "load-balancer" (really a desktop PC running balance-1.3.0)
> splitting requests over three machines. If you dive into the javascript
> you can figure out it's ip.
>
> 3) three back-end "geoserver boxes", each dual Xeon 2.8Ghz with a gig
> or a gig-and-a-half of ram. Each runs squid on port 80, doing
> http-acceleration of all HTTP-get requests from tomcat on port 8080.
> Each cache is configured to try and find the others for already cached
> URLS/tiles. These are not publically accessible via TCP/IP.
>
> 4) A beefy back-end SDE database, (also inaccessible to the outside
> world), dual opeteron 252's and 8GB RAM.
>
>
>
> As I poke around the map, I'm noticing all kinds of little bugs.
> Sometimes tiles just don't appear correctly (leaving little "holes" in
> the map at certain scales) and there seems to be some sort of text
> artifacts that indicate my SLD isn't correctly formed.
>
> Feel free to poke around. Places where people have already visited (and
> hence, their tiles have been cache in squid) will be licked-split.
> Other areas where people have not visited in 2400 seconds will not be so
> fast, as geoserver needs to render each tile separately.
>
>
> I'll get a clearer description of how to do this onto the wiki when I
> get a spare moment. Hopefully in a week or so.
>
> --saul
>
Cheers
Andrea Aime
_______________________________________________
User-friendly Desktop Internet GIS (uDig)
http://udig.refractions.net
http://lists.refractions.net/mailman/listinfo/udig-devel
--
Chris Holmes
The Open Planning Project
thoughts at: http://cholmes.wordpress.com
begin:vcard
fn:Chris Holmes
n:Holmes;Chris
org:The Open Planning Project
adr:;;377 Broadway, 11th Floor;New York;NY;10013;USA
email;internet:[EMAIL PROTECTED]
title:VP, Strategic Development
x-mozilla-html:FALSE
url:http://topp.openplans.org
version:2.1
end:vcard
_______________________________________________
User-friendly Desktop Internet GIS (uDig)
http://udig.refractions.net
http://lists.refractions.net/mailman/listinfo/udig-devel
